A major goal for biomedical researchers has been to repair damaged cells and tissues by reinitiating the body's developmental mechanisms. In this respect, the use of embryonic stem cells — and of induced pluripotent stem cells, which are generated by reprogramming differentiated adult cells — has seemed promising. But these cells come with various problems, including ethical concerns, and potential tumorigenicity and rejection by the host immune system. Three papers in this issue1,2,3 describe the direct conversion of neurons from skin fibroblast cells. There is hope, although no compelling evidence, that neurons generated in this way might be superior to those generated from induced pluripotent stem cells, thereby sidestepping the problems of using such cells4.
The conversion of mouse fibroblasts into neurons was reported last year5. The same team (Pang et al.1, page 220) now find, however, that introducing the same three transcription factors (Brn2, Ascl1/Mash1 and Myt1l) used in their original study into human fibroblasts is not sufficient to convert the cells into functional neurons; the resulting cells are immature, and even after 3 weeks in culture do not become functional, as judged by measurements of action potentials. Nonetheless, the authors show that the efficiency of human fibroblast conversion into neurons is enhanced by including an extra transcription factor, NeuroD1, a member of the bHLH family.
Caiazzo et al.2 (page 224) also set out to convert human fibroblasts into a specific type of neuron, in this case neurons of the midbrain that secrete the neurotransmitter dopamine. Starting their study in mice, they tested the efficiency of mixtures of transcription factors (combinations of Ascl1, Brn2, Myt1l and others) in the conversion process. The winning combination was that of Ascl1, Nurr1 and Lmx1a; Brn2 and Myt1l did not lead to dopamine-secreting neurons. The authors provide evidence that the conversion route was direct, and that the resulting neurons were functionally surprisingly similar to primary dopaminergic neurons isolated from the brain. What's more, when the converted dopaminergic cells were grafted into neonatal mouse brains, they became integrated and were functional.
Caiazzo and colleagues then applied their method to human fibroblasts, including those from patients with Parkinson's disease, a neurodegenerative disorder in which dopaminergic neurons are damaged. The conversion efficiency of adult human fibroblasts was 10–20-fold lower than that of their mouse equivalents, and the resulting cells were less mature. This discrepancy points to species differences between mice and humans2. The authors' observations fit well with those of another team6 that recently reported the direct conversion of human fibroblasts into dopaminergic cells using a different combination of transcription factors.
Yoo et al.3 (page 228) approached the search for a suitable conversion method from a different angle. They show that combined expression of miR-9/9* and miR-124 — two members of a class of short regulatory RNA sequences called microRNAs (miRNAs) — is sufficient to convert human fibroblasts into neurons. Again, the conversion process was greatly enhanced by the introduction of another bHLH-family transcription factor, NeuroD2. And increased expression of Ascl1 and Myt1l further increased the conversion efficiency of neonatal human fibroblasts: about 80% of the resulting neurons showed action-potential-like activity. As in the other two studies, Yoo and colleagues write that adult fibroblasts were less amenable to conversion.
Several common themes emerge. The studies1,2,3 highlight the importance of the bHLH-family transcription factors and the role of miRNAs in the conversion process. Moreover, they all show that the conversion of mouse and human fibroblasts into neurons requires different protocols, and that conversion of fibroblasts from adult or aged individuals is much less productive than that of embryonic or neonatal fibroblasts.
The essential role of NeuroD1 and NeuroD2 for efficient conversion correlates well with the crucial part played by these two transcription factors in neuron differentiation during normal brain development, and their reduced expression as neurogenesis decreases during ageing7,8. A relevant question is whether the degree of methylation of the genomic region containing the genes for these transcription factors is higher in human than in mouse fibroblasts and increases during ageing, thereby resulting in their reduced expression. DNA methylation is a crucial regulator of gene expression.
How miR-9/9* and miR-124 substitute for Brn2 — and possibly for Ascl1 and Myt1l — in the conversion process is not fully understood. Remodelling of chromatin (DNA–protein complexes), which in turn affects gene expression, seems to have a central role. Yoo et al.3 report that the two miRNAs modify the composition of the BAF chromatin-remodelling complex by regulating expression of its subunits, thereby transforming it into a chromatin remodeller that is characteristic of differentiated neurons9. But it remains unclear whether the specific chromatin structure in the converted neurons is the same as that in primary neurons.
In this respect, Caiazzo and colleagues' data2 are informative. They show that the gene-expression profiles of the dopaminergic neurons generated and those of isolated mouse midbrain dopaminergic neurons are similar but distinct: about 160 genes were expressed differently, with a more than fivefold variation in expression. This indicates that functional similarity of the converted neurons does not necessarily correspond to similar chromatin structure or gene-expression levels, and so cautions against the cells' premature clinical use. Instead, the differences between the converted neurons and their corresponding primary neurons must be characterized, and whether they give rise to unwanted side effects should be explored.
The miRNAs used by Yoo and co-workers3 probably also affect the expression of other proteins involved in cell-fate switching and neuron differentiation, including components of other chromatin-remodelling complexes. So before the clinical use of converted cells can be contemplated, further work should determine how the various chromatin-remodelling factors, and other factors that affect gene expression, contribute to cell conversion and how they can be controlled. Another question to be addressed is how the similarities and differences between converted cells and primary neurons, in terms of gene expression and chromatin structure, correlate with the functionality of the converted neurons — with their neurotransmitter production, firing of action potentials and functional integration into neuronal networks. It is also not known how gene expression and chromatin structure are shaped by intrinsic mechanisms and by the cells' immediate environment, in particular after their transplantation.
Thus, an area for further exploration is how a diseased brain's environment influences the functionality and gene-expression profiles of the transplanted converted neurons. Whether the converted cells are transplanted into the supportive environment of a neonatal mouse brain or into a diseased or aged human brain probably makes a difference. The finding that DNA methylation is dynamic in postnatal neurons10 raises hopes that the environment into which these converted cells are transferred could contribute to correctly shaping their gene-expression profile for long-term integration and function. However, the diseased brain in aged humans might not be so good at doing this. Therefore, investigating strategies to compensate for this deficit is perhaps the next big challenge in developing cell-based therapies for neurodegenerative disorders.